AllowedValues from botocore during --update-specs

[PatMyron]

most commonly maintained concise constraint format last time I checked:

src/cfnlint/data/ExtendedSpecs/all $ git grep 'AllowedValues":' | wc -l
AllowedValues: 187 # only 42 hardcoded AllowedValues lists remaining after this PR
NumberMin/NumberMax: 46
AllowedPatternRegex: 12
StringMin/StringMax: 8
ListMin/ListMax: 4
JsonMax: 2

#50 (comment), terraform-linters/tflint#274, terraform-linters/tflint#421

---

CloudSpecs is autogenerated by cfn-lint --update-specs

---

AllowedValues were hardcoded in ExtendedSpecs/all and patched into all the Resource Specifications in CloudSpecs when cfn-lint --update-specs was run. Now they're references to botocore types in ExtendedSpecs/all and cfn-lint --update-specs pulls from botocore directly to write that part of CloudSpecs

---

AWS::WorkSpaces::Workspace.ComputeTypeName should pick up POWERPRO and GRAPHICSPRO

AWS::CodeBuild::Project.Environment.Type should pick up WINDOWS_SERVER_2019_CONTAINER

AWS::S3::Bucket.TopicConfiguration.Event should pick up s3:ObjectRestore:* and s3:Replication:*

EbsVolumeType should pick up io2 from here

AWS::GuardDuty::IPSet.Format should pick up ALIEN_VAULT and PROOF_POINT and FIRE_EYE

AWS::CodeBuild::Project.Environment.ComputeType should pick up BUILD_GENERAL1_2XLARGE

AWS::CloudFront::Distribution.SslSupportMethod should pick up static-ip

AWS::WAFv2::RuleGroup.Rate.AggregateKeyType should pick up FORWARDED_IP

AWS::Glue::Connection.ConnectionInput.ConnectionType picks up NETWORK

AWS::EC2::LaunchTemplate.TagSpecification.ResourceType should pick up 42 values

---

Of the decent sized hardcoded AllowedValues lists left, AWS::SecretsManager::SecretTargetAttachment.TargetType, AWS::DMS::Endpoint.EngineName, AWS::Glue::Trigger.Condition.State, and AWS::AmazonMQ::Broker.EngineVersion are the only ones that changed in 2020. Will watch their documentation with http://followthatpage.com/

[kddejong]

Another option here is

Suggested change

	botocore_cache[service] = json.loads(get_url_content('https://raw.githubusercontent.com/boto/botocore/master/botocore/data/' + service + '/service-2.json'))
	botocore_cache[service] = json.loads(pkgutil.get_data('botocore', os.path.join('data', service, 'service-2.json')))

While we don't directly rely on botocore but it is getting loaded because of the SAM dependencies.

[PatMyron]

This is the biggest decision I've been debating
This would speed --update-specs up significantly, but would grab outdated information if botocore is outdated

$ time cfn-lint --update-specs # pulling from botocore latest on Github 
real  2m15.260s

$ time pip3 install botocore --upgrade
real	0m15.649s
$ time cfn-lint --update-specs # pulling directly from local copy of botocore
real	0m24.975s


# might be able to pull latest data faster than these individual requests though

$ time git clone https://github.com/boto/botocore/ --depth 1
real	0m7.271s

$ time curl -s -L https://codeload.github.com/boto/botocore/zip/master > botocore.zip
real	0m2.605s
$ time unzip botocore.zip
real	0m2.074s

https://stackoverflow.com/questions/56170380/downloading-files-raw-githubusercontent-com-is-immensely-slow

[miparnisari]

have you ensured that there is 0 difference in the json files before this PR and with this PR?

[PatMyron]

have you ensured that there is 0 difference in the json files before this PR and with this PR?

the difference in CloudSpecs is that all of these AllowedValues are now sorted and all the additional AllowedValues from botocore we were missing listed in the description